The right fit: SOVA models reduce dimensional errors

An automotive body at an optical measurement station (43K JPG)

A 3-D simulation of a manufacturing station (27K JPG)

Each year's new car models sport fancier features and consumer price tags ranging from $10,000 to more than $100,000. For automakers, a redesigned model might cost manufacturers billions of dollars — mostly for prototypes and new tooling. Assistant Professor Darek Ceglarek's method can help manufacturers reduce the number of costly prototypes and shorten new-product launch. Called a stream-of-variation-analysis (SOVA) model, it can reduce expensive dimensional errors during the design and production ramp-up phases.

Already, GM and DaimlerChrysler have implemented his methodology. Eventually, the SOVA models will become comprehensive software packages through which manufacturers in a variety of industries can simulate, predict and then eliminate errors both in a product's design and in its anticipated production process. Ceglarek's research will lay the theoretical foundation for reconfigurable and reusable assembly systems.

In addition, Ceglarek is collaborating with researchers at Texas A&M University and Motorola to develop methods to analyze and optimize a distributed sensor system in automotive and electronics assembly processes. The methods will help manufacturers install sensors in production lines to track and measure errors as they occur.